Research On The Capacity And Delay Of Cognitive Radio AD HOC Networks Based On Stochastic Geometry Theory

Cognitive radio technology is a promising approach to resolve the contradiction between spectrum shortage and low spectrum efficiency, which allows secondary users to access the authorized spectrum dynamically without interrupting the operation of the primary users. Cognitive radio ad hoc networks (CRAN) are the combination of cognitive radio technology and wireless Ad Hoc networks, which thus possess the advantages of high spectrum efficiency and self-orgnized network architecture. Performance characterization of CRAN must consider the distributed multi-hop routing structure, dynamic network topology and the spectrum availability varing with time and location.It is crucial to define reasonable and efficient performance metrics as measures for improving the network performance. In this dissertation, the capacity and delay of CRAN are studied based on stochastic geometry theory. The main contributions of this dissertation are listed as follows:1. Capacity and delay of CRAN are analyzed based on underlay spectrum sharing approach. Primary and secondary networks are modeled as two independent Poisson point processes (PPPs), respectively. Secondary users deploy the concurrent transmissions with primary users by limiting their density to meet the outage constraint of the primary network. The expected density of progress, defined as the product of the number of simultaneously successful transmissions per unit area and the average transmission distance per hop, is analyzed by adopting a selection region based routing protocol. In the protocol, the intermediate relay receivers must be located within a selection region which is jointly determined by a reference transmission distance and a directional angle. The expected densiy of progress is optimized seperately as well as optimized jointly with consideration of the reference transmission distance and the transmission probability. As a result, an optimal transmission probability and an upper bound of the optimal reference transmission distance are derived to maximize the expected density of progress. The closed-form results of the local delay and the end-to-end delay for CRAN are then derived.2. The capacity and the delay of CRAN are analyzed based on overlay spectrum sharing approach. The licensed channel occupancy of the primary network is modeled as a continuous-time two-state Markov process. Secondary network is modeled as a PPP. Slotted ALOHA MAC protocol is employed by secondary users to access the licensed channel when primary users are idle. By considering three sorts of relay receiver selection strategies, i.e., farthest, nearest and random relay selection, the closed-form expressions of the expected density of progress are derived. Numerical results show that the performance of the nearest relay receiver selection strategy outperforms that of the other two relay receiver selection strategies. Analytical expression of local delay for CRAN, defined as the mean number of time slots for one hop success propagation, is given subject to the constraint of the primary channel availability. The end-to-end delays of these relay receiver selection strategies are also derived. Numerical and simulation results show that a smaller end-to-end delay could be obtained by using nearest relay receiver selection routing strategy when transmission probability is small.3. The end-to-end delay of CRAN is analyzed by taking two traffic arrival models into account, i.e., the backlogged and geometric arrival. Primary network is modeled as a homogeneous PPP, while the secondary network is composed of many line multi-hop routes. A combined TDM A and ALOHA MAC protocol is used along each route in secondary network. Since queuing delay is considered, the conditions for stable queues is analyzed as well as the packet success probability and the density limits of secondary users for backlogged and geometric arrival models are presented, respectively. Then the local delay and the end-to-end delay are evaluated. Finally, the range of the optimal hop numbers and a equation with which the optimal transmission probability is satisfied are obtained to minimize the end-to-end dalay for backlogged arrival model and the equation of the optimal hop numbers for geometric arrival model..